What is ChatGPT actually good at?

Everyone uses it. Almost no one uses it well. Here's the fuller answer — the four things ChatGPT is genuinely good at, the catch nobody should skip, and how to prompt it properly.

Watch the episode

The basics

ChatGPT is a chatbot from OpenAI. You ask it something, and it writes back in plain language — answers, drafts, explanations. Underneath, it's predicting helpful text from its training. That one fact explains both its strengths and its biggest weakness.

Four things it's genuinely good at

1. First drafts. Emails, messages, essays, ideas — anything where a blank page is the enemy. Let it draft, then edit it into your voice.
2. Explaining hard topics simply. Ask it to explain something like you're new to it, then ask it to go one level deeper. That's a study loop, not a single answer.
3. Summarising long documents. Paste in a long article or report and ask for the key points, or for what it leaves out.
4. Helping you write and fix code. It can draft functions, explain error messages, and suggest fixes — but always test what it gives you before shipping it.

The catch

ChatGPT can sound completely confident while being wrong. It predicts plausible text, and plausible isn't the same as true. So don't trust it alone for medical, legal, or money decisions — and verify anything that matters before acting on it.

The prompt playbook

1. Give it context. Who it's writing for, what it's for, and what good looks like.
2. Give it an example. One example of the tone or format you want beats a paragraph of description.
3. Set constraints. Length, format, audience, and what to leave out.
4. Ask it to show its working. Step-by-step reasoning is easier to check than a bare answer.
5. Verify the result. Especially names, dates, quotes, and anything you plan to publish or act on.

Vague questions get vague answers. Specific prompts get useful ones.

Questions people ask next

Our takeaway

Sources & further reading